Replication is broken: Slave_IO_Running: No, Slave_SQL_Running: No

[artem-dronov]

Good day! gh-ost cannot find the replication master, previously we promoted the slave to the master, that is, if we SHOW SLAVE STATUS on the master, it will show Slave_IO_Running: No, Slave_SQL_Running: No, what should we do in this case?

./gh-ost --host=master_host --user=test_user --password=????? \ --database=test_db --table=test --assume-rbr --allow-on-master \ --verbose --alter="add test SMALLINT NULL" --chunk-size=3000 --assume-rbr \ --max-load=Threads_connected=20 > gh.log &

2021-06-02 09:08:53 INFO starting gh-ost 1.1.1 2021-06-02 09:08:53 INFO Migrating test_db.test 2021-06-02 09:08:53 INFO inspector connection validated on master_host:3306 2021-06-02 09:08:53 INFO User has REPLICATION CLIENT, REPLICATION SLAVE privileges, and has ALL privileges on test_db.* 2021-06-02 09:08:53 INFO binary logs validated on master_host:3306 2021-06-02 09:08:53 INFO Inspector initiated on slave_host:3306, version 5.7.32-log 2021-06-02 09:08:53 INFO Table found. Engine=InnoDB 2021-06-02 09:08:53 INFO Estimated number of rows via EXPLAIN: 75277111 2021-06-02 09:08:53 INFO Recursively searching for replication master 2021-06-02 09:08:53 INFO Tearing down inspector 2021-06-02 09:08:53 FATAL Replication on master_host:3306 is broken: Slave_IO_Running: No, Slave_SQL_Running: No. Please make sure replication runs before using gh-ost.

[shlomi-noach]

You should run RESET SLAVE ALL on your current primary. This is assuming there's nothing in the relay logs that you need. that's the normal thing to do after promoting a new primary -- but that's just my suggestion, make sure you understand what you are doing.

[artem-dronov]

You should run RESET SLAVE ALL on your current primary. This is assuming there's nothing in the relay logs that you need. that's the normal thing to do after promoting a new primary -- but that's just my suggestion, make sure you understand what you are doing.

Does this action not affect the health of the primary node?

[shlomi-noach]

right now your primary node thinks it's a replica. But if it's a primary, it probably shouldn't think so, it's risky. Because it may try to reconnect to whatever master_host is listed right now.

Define "affect the health"? RESET SLAVE ALL does not affect production traffic and is a non-blocking operation.

[artem-dronov]

I'm sorry, I haven't done this before. I'm more worried about whether the execution of this command on the master node will affect the synchronization of other slave nodes connected to the master.

[shlomi-noach]

Running this command on the primary node (master) does not affect the synchronization of the replicas connected to the primary node.

Obviously do not run it on the replicas.